1. Field of the Invention
This invention relates generally to three-dimensional (3D) imaging and more particularly to a device and method for capturing 3D images and video using a camera having a single lens barrel.
2. Description of the Related Art
Non-contact three-dimensional cameras, or digitizers, generally fall into four categories: stereoscopic digitizers, silhouette digitizers, timing digitizers, and projected pattern digitizers.
Stereoscopic digitizers traditionally employ multiple two-dimensional (2D) cameras to produce multiple viewing angles to capture multiple images of the target object from different angles. A 2D camera is positioned at a known offset relative to other 2D cameras. Given the positions of each camera it is possible to provide a correlation algorithm the necessary variables to identify the three-dimensional location of objects in the images.
Stereoscopic digitizers attempt to mimic the visual and mental facilities of the eyes and brain to identify the location of object surfaces in 3D space. The eyes 20 and brain 25 work in conjunction to obtain a three-dimensional mental model of the target object 5 (FIG. 1). Each eye 20 captures its own view (10a and 10b) and the two separate images which are processed by the brain 25. Each eye 20 has a slightly different placement, resulting in a different point of view and field of view 10a (left) and 10b (right) of the target object 5. As a result, each eye obtains a slightly different left image 15a and right image 15b of the target object 5. When the two images 15a and 15b arrive simultaneously in the back of the brain, they are united into one model, by matching up the similarities and adding in the small differences. Using the two images 15a and 15b, the brain compares the right image 15a and left image 15b to identify the number and magnitude of the similarities between the images to correlate the relationship between the images. Using the correlation between the images, the brain creates a 3D model of the target object 5.
A minimum requirement for stereoscopic digitizers is the ability to obtain two images from two different points of view. FIG. 2 illustrates a conventional 3D stereoscopic camera setup. Conventionally, obtaining the minimum two images is done with two distinct 2D cameras 50a and 50b, each positioned at a pre-defined distance from one another. Each 2D camera 50 includes an image pickup device, such as a CCD 30 and lens 35 positioned along an optical axis 40. Each camera 50 is positioned to point to the same target object 45.
By using an algorithm to identify the similar surfaces in the image obtained from camera 50a and camera 50b, and given the pre-defined distance between the cameras 50, the algorithm computes the three-dimensional location of the surface of target object 45.
One problem with stereoscopic digitizers is that they are generally both bulky and expensive because they require the use of multiple 2D cameras. Furthermore, the performance of the 3D camera setup is dependent on the careful configuration and alignment of the 2D cameras. Any change in the distance between the cameras or the angle between the cameras can pose problems to the pattern recognition algorithm, forcing the re-calibration of the hardware and software for the changed positions.